Evidence for Late-Time Feedback from the Discovery of Multiphase Gas in a Massive Elliptical at z=0.4
Fakhri S. Zahedy, Hsiao-Wen Chen, Erin Boettcher, Michael Rauch, K. Decker French, Ann Zabludoff
DD RAFT VERSION O CTOBER
28, 2020
Preprint typeset using L A TEX style AASTeX6 v. 1.0
EVIDENCE FOR LATE-TIME FEEDBACK FROM THE DISCOVERY OF MULTIPHASE GAS IN A MASSIVEELLIPTICAL AT z = 0 . F AKHRI
S. Z
AHEDY , H SIAO -W EN C HEN , E RIN B OETTCHER , M ICHAEL R AUCH , K. D ECKER F RENCH , AND A NN I. Z
ABLUDOFF The Observatories of the Carnegie Institution for Science, Pasadena, CA 91101, USA; [email protected] Department of Astronomy & Astrophysics, The University of Chicago, Chicago, IL 60637, USA Department of Astronomy, University of Illinois Urbana-Champaign, Urbana, IL 61801, USA Department of Astronomy & Steward Observatory, University of Arizona, Tucson, AZ 85721, USA
ABSTRACTWe report the first detection of multiphase gas within a quiescent galaxy beyond z ≈ . The observations usethe brighter image of doubly lensed QSO HE 0047 − M star ≈ M (cid:12) )elliptical lens galaxy at z gal = 0 . . Using Hubble Space Telescope ’s Cosmic Origins Spectrograph (COS),we obtain a medium-resolution FUV spectrum of the lensed QSO and identify numerous absorption featuresfrom H in the lens ISM at projected distance d = 4 . kpc. The H column density is log N (H ) / cm − =17 . +0 . − . with a molecular gas fraction of f H = 2 − , roughly consistent with some local quiescent galaxies.The new COS spectrum also reveals kinematically complex absorption features from highly ionized speciesO VI and N V with column densities log N (O VI ) / cm − = 15 . ± . and log N (N V ) / cm − = 14 . ± . ,among the highest known in external galaxies. Assuming the high-ionization absorption features originate ina transient warm ( T ∼ K) phase undergoing radiative cooling from a hot halo surrounding the galaxy, weinfer a mass accretion rate of ∼ . − . (cid:12) yr − . The lack of star formation in the lens suggests the bulkof this flow is returned to the hot halo, implying a heating rate of ∼ erg yr − . Continuous heating fromevolved stellar populations (primarily SNe Ia but also winds from AGB stars) may suffice to prevent a largeaccumulation of cold gas in the ISM, even in the absence of strong feedback from an active nucleus. INTRODUCTIONHow and why some galaxies cease forming stars and re-main quiescent are open questions that bear significantly onour understanding of galaxy evolution. Contrary to the ex-pectation that a lack of star-formation is the consequenceof a paucity of cool gas, observational studies have estab-lished that a high fraction of passive galaxies are not gas-poor(see Chen 2017a and references therein). Systematic 21cmsurveys have discovered that more than a third of present-day quiescent galaxies contain abundant neutral hydrogen(H I ) gas in their interstellar medium (ISM; e.g., Oosterlooet al. 2010; Serra et al. 2012). At an earlier epoch, QSOabsorption-line surveys of Mg II absorption features near lu-minous red galaxies (LRGs) at z ∼ . have also demon-strated that a significant fraction of these distant massive el-lipticals (with total stellar masses of M star (cid:38) M (cid:12) ) aresurrounded by chemically enriched cool gaseous halos on ∼ kpc scales (e.g., Gauthier et al. 2009, 2010; Bowen &Chelouche 2011; Huang et al. 2016; Chen et al. 2018). Thetotal mass in this cool ( T ∼ K) circumgalactic medium(CGM) is estimated to be M cool ≈ (1 − × M (cid:12) within projected distance d < kpc (or as much as ≈ × M (cid:12) at d < kpc; Zahedy et al. 2019), simi-lar to what has been reported for star-forming galaxies (e.g.,Chen et al. 2010; Stocke et al. 2013; Werk et al. 2014).The existence of large reservoirs of cool gas around mas-sive ellipticals challenges simple theoretical expectations that these galaxies are surrounded by predominantly hot ( T (cid:38) K) gas on both small ( (cid:46) kpc; ISM) and large ( ∼ kpc; CGM) scales. Furthermore, it indicates that some physi-cal mechanisms are preventing the gas from triggering the re-sumption of star formation in the central galaxy. A commonfeature of the gaseous environment at d (cid:46) kpc aroundmassive quiescent galaxies is the high Fe / Mg abundance ra-tio, [Fe / Mg] > ∼ , that has been observed in every instancecool gas is present (Zahedy et al. 2016, hereafter Z16; Za-hedy et al. 2017a). This Fe enhancement not only indicatesthat the ISM of massive ellipticals has been significantly en-riched by Type Ia supernovae (SNe Ia), but also points to SNeIa as a potentially important heating source in massive halos(e.g., Conroy et al. 2015; Li et al. 2020a,b).One of the galaxies studied in Z16 is a massive ( M star ≈ M (cid:12) ) elliptical lens for QSO HE0047 − z gal =0 . ± . . It exhibits extremely strong and kinemati-cally complex low-ionization metal absorptions with a line-of-sight velocity spread exceeding 600 km s − (Figure 1,top) and a velocity shear of ≈ km s − between two lo-cations ≈ kpc apart in projection. Long-slit far-ultraviolet(FUV) spectroscopic observations of both lensed QSO im-ages revealed the presence of abundant H I within the galaxy,with measured H I column densities of log N (H I ) / cm − =19 . − . at both locations (Zahedy et al. 2017b, here-after Z17), constraining the gas metallicity to be [Fe / H] (cid:38) for both sightlines after accounting for likely dust depletion.While Z17 also noted the presence of possible absorption fea- a r X i v : . [ a s t r o - ph . GA ] O c t Z AHEDY ET AL .tures from other metal ions probing a wide range of ioniza-tion states, including the highly ionized O VI λλ , doublet, their low-resolution spectra precluded a detailed in-vestigation of these absorption profiles to confirm the pres-ence of high ions. Because O ions are most abundant attemperatures near the peak of the cooling curve for metal-enriched gas ( T ≈ . K; e.g., Gnat & Sternberg 2007),such warm gas is expected to cool rapidly if left to it-self. Therefore, the possible detection of rapidly coolinggas in the ISM implies the presence of an effective heat-ing mechanism in the galaxy. Characterizing the propertiesof such a transient gas phase and its relationship to cooleratomic/molecular gases offers a unique opportunity to under-stand the dynamic gas content of ellipticals, in order to gaininsight into late-time feedback in massive quiescent galaxies.In this
Letter , we report the robust detection of highlyionized gas in the ISM of the massive elliptical lens ofHE0047 − VI and N V absorption fea-tures. Furthermore, we report the serendipitous discoveryof molecular hydrogen ( H ) in the ISM, the first direct de-tection of H within a passive galaxy beyond the local Uni-verse. We compare the spatial distributions and mass budgetsof the molecular ( T ∼ K), cool ( T ∼ K), and warm( T ∼ K) ISM phases and discuss their implications forfeedback in massive ellipticals. We adopt a Λ cosmologywith Ω M = 0 . , Ω Λ = 0 . , and H = 70 km s − Mpc − . OBSERVATIONSNew FUV spectra of image A of the doubly lensed QSOHE 0047 − z QSO = 1 . ; Figure 1 of Z16) were ob-tained with the Cosmic Origins Spectrograph (COS) onboardthe Hubble Space Telescope (HST) during our
HST
Cycle25 observing program (Program ID: 15250; PI: Zahedy) inDecember 2018.
HST /COS with the G130M and G160Mgratings provides a wavelength coverage from λ ≈ ˚A to λ ≈ ˚A at a medium resolution of FWHM ≈ − km s − , a fifteenfold increase in resolution from theZ17 spectra. The total integration time of the observationswas 9,418 s and 17,722 s for the G130M and G160M grat-ings, respectively, comprising 22 individual exposures spreadover three separate HST visits. The observations used two(four) central wavelength settings for the G130M (G160M)grating and two or four FP-POS at each central wavelength,to ensure a continuous wavelength coverage and reduce fixedpattern noise over the full spectral range of the data.The pipeline-reduced COS data were downloaded from the
HST archive and processed further using our custom soft-ware. The additional data reduction involved recalibratingthe COS wavelength solution using a method described inChen et al. (2018) and Zahedy et al. (2019). These stepsresulted in a combined spectrum which was then contin-uum normalized by fitting a low-order polynomial function toabsorption-free spectral regions. The final COS spectrum ofHE 0047 − A has a median signal-to-noise ratio of S/N ≈ − per resolution element over the full wavelengthrange. The wavelength solution is accurate and precise tobetter than 3 km s − , as evidenced by a comparison betweenlow-ionization absorption features seen in COS and ground- FeII 2600 -1000 -500 0 500 1000 1500
Relative Velocity (km/s)
HI 1215HI 1215 log N (HI) / cm − =19 . ± . N o r m a li z e d F l u x Figure 1 . Top : Kinematically complex gas at d = 4 . kpc from themassive elliptical lens galaxy ( z gal = 0 . , vertical dashed line),seen in Fe II λ absorption from ground-based optical echellespectrum of HE 0047 − A (adapted from Z16). The absorptionprofile comprises 15 individual components (blue tick marks) span-ning over km s − in line-of-sight velocity. Zero velocity cor-responds to the redshift of the H absorption identified in Figure 2, z abs = 0 . . Bottom : New
HST /COS FUV spectrum of thecorresponding Ly α absorption associated with the lens galaxy. TheCOS spectrum is rebinned by three pixels for display purposes. The1- σ error spectrum is included in cyan. Contaminating features aredotted out for clarity. The magenta tick mark above the profile in-dicates the best-fit centroid of the damped Ly α profile. The solidred and dashed magenta curves show the best-fit N ( H I ) and itsuncertainty, log N ( H I ) / cm − = 19 . ± . . based optical echelle spectra (presented in Z16) and the ex-cellent agreement in line centroids among various H absorp-tion lines spanning ≈ ˚A in observed wavelength (§3.1).We supplement our COS spectrum of sightline A with low-resolution ( FWHM ≈ km s − ) FUV spectra of both im-ages of the lensed QSO taken with the Space Telescope Imag-ing Spectrograph (STIS) and the G140L grating onboard HST from Z17. The STIS spectrum of HE 0047 − A ( B )has a median S/N ≈ −
30 (12 − per resolution elementover its full wavelength range of − ˚A. RESULTSA prominent feature associated with the massive ellipti-cal lens galaxy is the Ly α absorption with strong dampingwings (Figure 1, bottom), confirming the previously reportedhigh N ( H I ) of the gas inferred using low-resolution STISFUV spectra (Z17). To refine the N ( H I ) measurement,we perform a Voigt profile analysis on the observed dampedLy α profile using a custom software (see Zahedy et al. 2019)that takes into account the relevant COS line-spread function(LSF; Lifetime Position 4). Our analysis yields a total H I column density of log N ( H I ) / cm − = 19 . ± . , whichis consistent within uncertainties with the Z17 measurement.We adopt this N ( H I ) throughout subsequent analysis.3.1. Discovery of H in the ISM of the Lens Galaxy A visual inspection of our
HST /COS spectrum ofHE 0047 − A reveals the presence of numerous absorp-tion features consistent with the H Lyman and Werner bandsat redshift z ≈ . , or approximately − km s − from ULTIPHASE GAS IN A MASSIVE ELLIPTICAL GALAXY AT z = 0 . Table 1 . H properties at d = 4 . kpc from the lens galaxy z abs J log N/ cm − b (km s − ) . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +1 . − . . +0 . − . . +1 . − . . +0 . − . . +5 . − . < . a Total . +0 . − . a
95% upper limit (see §3.1). the systemic redshift of the lens galaxy (see Figure 2). The H absorption features coincide in velocity with the strongestlow-ionization absorption component identified in Z16 (com-ponent 1 in their Table 6). We are able to identify more than140 absorption transitions originating from the ground stateof H at different rotational levels from J = 0 to J = 5 .Each of these transitions has a vibrational quantum numberof ν = 0 for the lower state and ν ≤ ( ν ≤ ) for the upperstate in the Lyman (Werner) band.To characterize the molecular gas properties, we performa Voigt profile analysis using a custom software that modelsthe observed H transitions in each J level simultaneously.We adopt the H line list from Ubachs et al. (2019) whichwas made available to us by Patrick Petitjean (private com-munication). Although we detect more than 140 H transi-tions, a significant fraction of these lines are blended witheach other or other absorption lines. To ensure robust fittingresults, we perform our absorption analysis on a subset ofavailable lines (between five and 14 transitions) for each J value, which are selected to contain minimal blending andhave unambiguous local continuum level. While only a frac-tion of observed H transitions are used to find the best-fitmodel, we find that the resulting full H absorption modelreproduces the absorption profiles of most of the excludedtransitions reasonably well.For each rotational J level, we first generate a model spec-trum for a single-component line profile, motivated by boththe narrow linewidths and lack of kinematic substructures inthe observed H absorption profiles (see Figure 2). The Voigtprofile is uniquely defined by three free parameters: the linecentroid redshift z abs , the absorption column density log N ,and the Doppler parameter b . To reduce the number of freeparameters, all transitions from a given J level are tied tohave the same log N and b . We further require different J levels to share the same line centroid redshift. Once a theo- While the velocity offset of the H absorption features may seem largefor ISM gas, it is partly explained by the uncertainty on the lens redshift( ≈ km s − ; Z16). Furthermore, the projected escape velocity at r = 5 kpc from the lens galaxy is ≈ − km s − given the estimated massof its host dark-matter halo (Z16), so the observed H kinematics is consis-tent with ISM gas that is bound to the galaxy. Empirically, large kinematicwidths of ≈ km s − have been observed in the atomic/molecular ISMof some nearby early-type galaxies (e.g., Oosterloo et al. 2007; Davis et al.2013), reflecting the potential wells of these massive systems. retical H absorption spectrum has been generated, it is con-volved with the relevant COS LSF and subsequently binnedto match the pixel resolution of the data. Finally, this modelspectrum is compared to the data and the best-fit model pa-rameters for each J level are found by minimizing χ valueat the selected H transitions. We estimate the model un-certainties by constructing a marginalized posterior probabil-ity distribution for each model parameter based on a MarkovChain Monte Carlo (MCMC) analysis done with the E MCEE package (Foreman-Mackey et al. 2013). Each MCMC runconsists of 500 steps performed by an ensemble of 250 walk-ers, which are seeded in a small region of the parameter spacearound the minimum χ solution to speed up convergence.We present the best-fit model absorption profiles and com-pare them to the data in Figure 2. In addition, we sum-marize the results of the Voigt profile analysis in Table 1,where we report the model parameters and estimated 68%confidence intervals for the J = 0 to J = 5 levels. For J = 6 , which does not exhibit any detectable absorption, wereport in Table 1 the 95% upper limit on the absorption col-umn density for a b = 10 km s − line profile (matching thelinewidth of the J = 5 level), estimated using the error ar-ray at the strongest available J = 6 transition in the COSdata. The best-fit model yields a total H column densityof log N (H ) / cm − = 17 . +0 . − . and a best-fit redshift of z abs = 0 . ± . . The centroid of the H lineprofile is consistent within uncertainties ( < km s − ) withthe strongest low-ionization metal component identified inground-based optical echelle spectra (Z16), which indicatestheir association.As shown in Table 1, our analysis also identifies a trend ofincreasing Doppler parameter with increasing J value, from b ≈ km s − at J = 0 to b ≈ km s − at J = 5 . The trendof rising velocity dispersion with rotational level has beenreported in a number of H -bearing damped Ly α absorbers(DLAs) at low and high redshifts (e.g., Ledoux et al. 2003;Albornoz V´asquez et al. 2014; Boettcher et al. 2020). Thekinetic temperature needed to thermally broaden the H lineprofiles to a linewidth of ≈ km s − is ≈ (10 ) K,significantly higher than temperatures at which a significantamount of molecular gas is expected to be present. There-fore, our measurements indicate that non-thermal line broad-ening is dominant for both low and high J states in the gas,with increasing turbulence toward higher rotational states.We discuss the possible origins of this trend in §4.1.With both the neutral and molecular hydrogen contentsof the absorber known, we can estimate the moleculargas fraction according to the following expression, f H =2 N (H ) / [2 N (H ) + N (H I)] . The extremely strong low-ionization absorber observed along HE 0047 − A is re-solved into 15 kinematic components (Figure 3; Z16). Whilethe total N (H I ) can be measured robustly from the strongLy α damping wings (Figure 1), it is not possible to constrainthe H I column densities of these individual components be-cause all available H I Lyman series lines are heavily satu-rated and the different components blended. Thus, we firstestimate f H by attributing all the observed N (H I ) to the H -bearing component. Although this assumption is unreal- Z AHEDY ET AL . J = 0 J = 1 J = 2 J = 3 J = 4 J = 5 Relative Velocity (km/s) N o r m a li z e d F l u x J = 6 Figure 2 . Continuum-normalized absorption profiles of select H Lyman- and Werner-band transitions that are used in our absorption analysis,grouped by J level, observed along sightline HE 0047 − A at d = 4 . kpc from the massive elliptical lens. The COS spectrum is rebinnedby three pixels for display purposes. The 1- σ error spectrum is included in cyan. Zero velocity marks the best-fit redshift of the H absorptionidentified with a Voigt profile analysis, z abs = 0 . . Regions excluded from the analysis due to blending and/or contaminating featureshave been grayed out for clarity. The best-fit H absorption profiles are plotted on top of the data in red curves. The significant detection of H at J > indicates that non-thermal excitation mechanism is effective in populating these high rotational levels (see §4.1). istic because it would result in highly asymmetric Ly α damp-ing wings owing to the H -bearing component occurring atthe blue extremum of the profile, it yields a conservative lower limit on f H of log ( f H ) lower = − . +0 . − . . To esti-mate an upper bound on the molecular gas fraction, we notethat the H -bearing component contains ≈ − % of thetotal column densities of the low-ionization species probedby Mg I , Mg II , and Fe II absorptions (see Table 6 of Z16). Ifwe assume that all 15 components have similar metallicities and dust content, which is justified by the relatively uniform Fe / Mg elemental abundance ratio observed across all com-ponents (Z16), then the inferred N (H I ) of the H -bearingcomponent is log N ( H I ) / cm − ≈ . . Consequently, theimplied molecular gas fraction is log ( f H ) upper = − . +0 . − . .The observed f H at d = 4 . kpc from the lens galaxy iscomparable to nearby ellipticals with CO detections (Welchet al. 2010; Young et al. 2014) but is among the highestknown for z < DLAs, where ≈ of absorbers with ULTIPHASE GAS IN A MASSIVE ELLIPTICAL GALAXY AT z = 0 . L o w - i o n i z a t i o n g a s H i g h - i o n i z a t i o n g a s Figure 3 . Continuum normalized absorption profiles of differenthigh- and low-ionization metal transitions along HE 0047 − A at d = 4 . kpc from the massive elliptical lens. Zero velocity cor-responds to the redshift of the H absorption detected in Figure 2,whereas the systemic redshift of the lens galaxy, z gal = 0 . , isshown in vertical dashed line. The 1- σ error spectrum is included incyan. Contaminating features have been dotted out for clarity. Themagenta tick marks at the top of the first three panels indicate thelocation of individual components for the high-ionization speciesidentified in the Voigt profile analysis (see §3.2), with the best-fitVoigt profile models included in red. For comparison, individualcomponents of the low-ionization species are marked with the bluetick marks in the bottom five panels (Z16). The high ions show adistinct kinematic structure from what is seen in the low ions, indi-cating that they arise in a different gas phase. log N ( H I ) / cm − (cid:38) have log f H (cid:46) − (e.g., Crightonet al. 2013; Muzahid et al. 2015a; 2016; but see Boettcheret al. 2020). Considering that H forms on the surface ofdust grains, the high f H can be explained by the high gasmetallicity, [Fe / H] (cid:38) (Z17), which results in an elevateddust-to-gas ratio relative to the general DLA population.3.2. Highly Ionized Gas in the ISM of the Lens Galaxy
Z17 previously noted possible absorption features fromhigh-ionization metal lines associated with the lens galaxyalong both sightlines of HE 0047 − Table 2 . High-ionization absorption properties at d = 4 . kpc Component Species d v ca b log N / cm − (km s − ) (km s − )1 O VI +32 . ± . . ± . . ± . N V . ± . VI +220 . ± . . ± . . ± . N V . ± . VI +372 . ± . . ± . . ± . N V . ± . VI +423 . ± . . ± . . ± . N V . ± . VI +507 . ± . . ± . . ± . N V . ± . VI +672 . ± . . ± . . ± . N V . ± . a Relative velocity shift from the H absorption redshift, z abs = 0 . The
HST /COS spectrum of HE 0047 − A clearly resolvesdifferent metal absorption profiles, enabling precise measure-ments of gas kinematics and column densities of the highlyionized species.As shown in the top three panels of Figure 3, the new COSspectrum confirms that O VI absorption is indeed detectedand resolved into multiple components in the lens galaxy. Inaddition, N V absorption is also detected with a kinematicstructure that is consistent with O VI . To constrain their ab-sorption properties, we perform a joint Voigt profile analysisof the O VI λ line and the N V λλ , doubletfollowing the method of Zahedy et al. (2019). To ensure therobustness of the fit, we tie both the component structure andDoppler linewidths of the two ions. We summarize the re-sults from our Voigt profile analysis of these high-ionizationspecies in Table 2. The continuum-normalized absorptionprofiles and best-fit models for O VI and N V are presentedin the top three panels of Figure 3. To compare the kinemat-ics between low- and high-ionization species, we also showthe observed and modeled absorption profiles of Mg I , Mg II ,and Fe II in the bottom five panels of Figure 3, from previousabsorption analysis reported in Z16.It is clear from Figure 3 that the high ions exhibit a dis-tinct kinematic structure from that of the low ions, whichindicates that the high ions arise in a different gas phase(Zahedy et al. 2019). Specifically, our analysis reveals ahighly ionized gas phase in the lens ISM that is kinemati-cally complex, comprising six broad kinematic components( b ≈ − km s − ) that span ≈ km s − in line-of-sight velocity. The observed total column densities of thesehighly ionized species are log N (O VI ) / cm − = 15 . ± . and log N (N V ) / cm − = 14 . ± . . These O VI and N V The second member of the O VI doublet, O VI λ , is excludedfrom the Voigt profile analysis due to significant blending with neighboringlow-ionization transitions C II λ and O I λ . Although the O VI λ profile is also contaminated by a higher-redshift Ly (cid:15) line at z =0 . , in this case the absorption profile of the contaminating Ly (cid:15) line iswell-constrained by various other Lyman series lines observed in our COSspectrum. To remove this contamination from the O VI λ absorption,we have divided the observed O VI λ profile by the best-fit model ofthe z = 0 . Ly (cid:15) line prior to performing the analysis. Z AHEDY ET AL .absorbers are among the strongest known to be in the vicin-ity of z < galaxies (cf., Johnson et al. 2015; Muzahid etal. 2015b; Werk et al. 2016; Rosenwasser et al. 2018; Za-hedy et al. 2019), where high-ionization absorbers with log N (O VI ) / cm − > and log N (N V ) / cm − > are rare.It is also interesting to note the observed N V to O VI column density ratios among the six high-ionization com-ponents, which have an arithmetic mean and dispersion oflog (cid:104) N (N V ) /N (O VI ) (cid:105) = − . ± . . These ionic ra-tios are considerably higher than typical values seen in theGalactic corona (e.g., Wakker et al. 2012), the circumgalac-tic medium of external galaxies (e.g., Werk et al. 2016; Za-hedy et al. 2019), and the high-redshift intergalactic medium(e.g., Lehner et al. 2014), where a large majority of absorbersin these diverse environments exhibit log N (N V ) /N (O VI ) (cid:46) − . . We argue that a super-solar [N / O] in the highlyionized gas phase is the most likely explanation, consideringthat high nitrogen-to-alpha ratios of [N /α ] (cid:38) . have beenreported in the evolved stellar populations and cool ISM ofnearby ellipticals (e.g., Greene et al. 2013; Yan 2018). Simi-lar [N / O] ratios in both high- and low-ionization gases wouldalso suggest a causal link between different phases of theISM of the elliptical lens (we discuss this connection in §4.3). DISCUSSION4.1.
Physical Conditions of the H Gas
The distribution of H molecules among different rota-tional levels reflects the excitation state of the gas and of-fers insight into the physical mechanisms that are responsi-ble. The relative populations of different H rotational lev-els can be described by a Boltzmann distribution, N J N J =0 = g J g J =0 exp[ − B v J ( J + 1) /T ] , where N J is the H columndensity for the rotational level J , T is the excitation tem-perature from J = 0 to rotational level J , and B v = 85 . K. The statistical weight g J is (2 J + 1) for even-numbered J or J + 1) for odd-numbered J . In Figure 4, we show the H excitation diagram of the absorber for rotational states be-tween J = 0 and J = 5 . The column density ratio between J = 0 and 1 states, which contain ≈ % of the total N (H ) ,implies an excitation temperature of T = 104 +39 − K.The observed T for the bulk of molecular gas along thelensed sightline is comparable to typical values reported in z < -bearing DLAs (see Muzahid et al. 2015a). How-ever, while the observed population for J = 2 can alsobe well-reproduced by the same excitation temperature, thissingle-temperature model fails to explain the observed col-umn densities at J > (Figure 4, left panel). The predictedcolumn densities for these higher rotational states are ordersof magnitude lower than the observations, which indicates ahigher excitation temperature for J > .It is well-known from Galactic H studies that a one-temperature fit typically works only for optically thin H absorbers with log N (H ) / cm − (cid:46) (e.g., Spitzer et al.1974; Spitzer & Jenkins 1975; Jenkins & Peimbert 1997).In contrast, stronger H absorbers in the Galaxy and beyondhave been found to exhibit elevated populations at higher ro-tational levels (e.g., Jenkins & Peimbert 1997; Reimers et al.2003; Noterdaeme et al. 2007; Rawlins et al. 2018; Balashev et al. 2019; Boettcher et al. 2020), which indicate that the H gas is bifurcated into two excitation temperatures. Motivatedby these prior observations, we perform a simultaneous fitof a two-temperature model to our data and find that the ob-served H level populations are well-reproduced by two exci-tation temperatures of T , = 93 +20 − K and T , = 490 +48 − K (Figure 4, middle panel).The observed temperature bifurcation and trend of risingvelocity dispersion with J level (see §3.1) can be understoodto be a consequence of the H absorption originating in a gascloud with an internal density and/or temperature stratifica-tions (e.g., Noterdaeme et al. 2007). In this scenario, mostof the column densities at low- J levels originate from the in-ner layer of the cloud, where the gas is sufficiently dense andshielded from radiation that collisions are the dominant exci-tation mechanism. Consequently, the low-level populationsare essentially thermalized and the lower excitation tempera-ture is highly coupled to the kinetic temperature of the gas.In contrast, the elevated column densities and broader lineprofiles of high- J levels indicate that they arise primarilyfrom warmer and more turbulent outer layers of the cloud(e.g., Lacour et al. 2005). At these locations, H moleculescan be highly excited through collisions triggered by shocksand turbulent dissipation (e.g. Jenkins & Peimbert 1997; Gryet al. 2002; Gredel et al. 2002; Ingalls et al. 2011), as well asthrough radiation pumping by an external UV radiation field(e.g., Jura 1975; Klimenko & Balashev 2020). A unique pre-diction of the shock scenario is a systematic shift of up to afew km s − in line centroids with increasing rotational state,which is caused by the different J levels originating from dis-tinct locations moving at slightly different speeds relative tothe shock front (e.g. Jenkins & Peimbert 1997; Gredel et al.2002). Although we do not detect any systematic shift in linecentroids with J levels to within the precision of our COSwavelength calibration ( (cid:46) km s − ), we cannot rule out amore modest shift of ≈ km s − or less, which may be theresult of weaker shocks (e.g., Gredel 1997).As an alternative, we now explore radiation pumping asan excitation mechanism. We perform a series of calcula-tions using the C LOUDY v.13.03 code (Ferland et al. 2013)to simulate a plane-parallel slab of gas with uniform density n H which is irradiated by two UV radiation fields: the up-dated Haardt & Madau (2001) extragalactic UV backgroundat z = 0 . , known as HM05 in C LOUDY , and the built-inunextinguished Milky Way ISM radiation field from Black(1987). To constrain the strength of UV radiation that is re-quired to reproduce the observations, we vary the overall in-tensity of the ISM radiation field by a scale factor of between0.1 and 100. We incorporate dust grains in the calculationsfollowing the observed grain abundance and size distributionin the local ISM. For each input radiation field, we construct agrid of C
LOUDY models spanning a wide range of gas densi-ties ( ≤ log n H / cm − ≤ ) at the observed gas metallicity(Z17). For each grid point, C LOUDY calculates the expectedcolumn density for each J level assuming thermal and ion-ization equilibrium. To simulate two-sided illumination ofthe cloud, we use half the observed N (H ) as the stoppingcondition for the calculations and subsequently double the ULTIPHASE GAS IN A MASSIVE ELLIPTICAL GALAXY AT z = 0 . J J J Figure 4 . Excitation diagram for the observed rotational level populations (red circles) of H gas at d = 4 . kpc from the massive elliptical lens. Left : Assuming that different rotational levels follow a Boltzmann distribution, the observed ratio between the J = 0 and J = 1 levels indicatesan excitation temperature of T = 104 +39 − K (dashed line). However, this single-temperature model severely underpredicts the observationsat
J > . Middle : A model with two excitation temperatures of T , = 93 +20 − K (dotted line) and T , = 490 +48 − K (dash-dotted line) canreproduce the observed level populations. The thin gray curves show 100 random realizations of the two-temperature model using the MCMCmethod.
Right : The thin blue curves represent a set the
CLOUDY models that best reproduce the trend seen in the data. These models have UVradiation fields which are − times more intense than the Milky Way ISM radiation field. If the elevated populations at higher rotationallevels are due to radiation pumping, the required radiation field is significantly higher than what is observed in the local Galactic ISM. output H level populations for comparison with the data.We summarize the results of our C LOUDY calculations inthe right panel of Figure 4, where the set of models that bestreproduce the observed H excitation diagram are shown inthin blue curves. These models have UV radiation fieldswhich are − times stronger than the local ISM radi-ation field. The range of gas densities are n H ≈ − − , with mean H kinetic temperatures ( − K)and H I column densities (log N ( H I ) / cm − = 19 . − . )which are broadly consistent with the observations. While itis clear that these simple models are only able to roughly re-produce the general trend seen at J > , this exercise demon-strates that if the elevated populations at higher rotational lev-els are primarily due to radiation pumping, the required UVradiation field is significantly higher than what is observed inthe Galactic ISM (see also e.g., Klimenko & Balashev 2020;Boettcher et al. 2020).4.2. Spatial Variations in Multiphase Gas Properties
A benefit of using a multiply lensed QSO system as gasprobes is the ability to investigate spatial variations in the gasproperties of a foreground galaxy. As described in Z16 (seetheir Figure 1), the doubly lensed images of HE 0047 − A at d = 4 . kpc (1.8 half-light radii, r e )and sightline B at d = 3 . kpc (1.3 r e ). The observed H I and low-ionization metal column densities differ by less than . − . dex between the two sightlines (Z16; Z17), despitea separation of ≈ kpc in projection. These similarities sug-gest that the cool ( T ∼ K) ISM phase is spatially ex-tended, with a high gas covering fraction at d (cid:46) kpc.While we are unable to perform a detailed analysis on theO VI absorption detected in the low-resolution STIS FUVspectrum of HE 0047 − B (Z17), we can compare the general absorption properties of the O VI absorbers detectedalong the two sightlines. Specifically, the total O VI rest-frame equivalent width is W r (1031) B = 1 . ± . ˚A alongsightline B , which is very similar to what is observed in theCOS spectrum along sightline A , W r (1031) A = 1 . ± . ˚A. Furthermore, the observed FWHM of the O VI profilealong sightline B is ≈ km s − , which is comparableto the observed kinematic spread of ≈ km s − for theO VI absorption profile along sightline A (§3.2). The co-herent O VI absorption properties between the two sightlinesimply that similar to the low-ionization gas phase, the highlyionized ISM is spatially extended and has a high coveringfraction on a scale of ∼ kpc in the massive elliptical.The lack of a high-resolution FUV spectrum ofHE 0047 − B prevents a direct search for H along thissightline. To assess whether we can constrain spatial vari-ations in molecular gas properties using the available low-resolution STIS FUV spectrum of sightline B , we per-form the following experiment. First, we divide the best-fitmodel for the full Lyman and Werner bands from the high-resolution COS FUV spectrum of sightline A to remove all H absorption from the spectrum. Then, we convolve the re-sulting “ H -free” spectrum of the QSO with the STIS LSFand compare the result with our STIS spectrum of sightline A . We find that while individual H lines are unresolved inthe STIS spectrum, the combined absorptions from the Ly-man and Werner bands result in an overall flux decrementthat is detectable across the QSO spectrum.Motivated by the result of the experiment, we generate aseries of model Lyman and Werner bands spanning a widerange of N (H ) , apply them to the “ H -free” QSO spectrum,and convolve the results with the STIS LSF. Each of the re-sulting spectra is then rescaled to the level of sightline B us-ing the mean observed flux ratio of the two lensed images in Z AHEDY ET AL .two absorption-free regions: − and − ˚A in the observed frame. Finally, we compare the productsto the STIS spectrum of sightline B in the spectral region be-tween 1415 and 1435 ˚A in observed wavelength, which has alarge concentration of strong H transitions, and infer the al-lowed N (H ) using a χ analysis. The observed spectrum ofHE 0047 − B is consistent with the presence of H with N (H ) (cid:46) cm − at the 95% confidence level.The inferred molecular gas fraction of f H (cid:46) . alongsightline B is a factor of at least ≈ − times lowerthan that observed along sightline A on the opposite side ofthe galaxy, f H = 2 − (§3.1). This exercise suggests thatin contrast to the neutral and highly ionized gas phases, themolecular gas distribution in the lens ISM is clumpier. Fur-thermore, the observed f H along the two lensed sightlinesare consistent with nearby quiescent galaxies found to har-bor molecular gas (e.g., Young et al. 2014) but low comparedto typical values in star-forming disks (Chen 2017b and ref-erences therein). If these f H constraints are representativeof the rest of the galaxy, they imply a low mass fraction ofdense, cold molecular gas in the multiphase ISM of the lens.4.3. Implications for Feedback in Massive Ellipticals
How the ISM is partitioned by mass into its different gasphases depends sensitively on the gas cooling rate, the avail-able heating to offset this cooling, and the relevant timescalesof these processes. The simultaneous detections of multiplegas phases in the lens ISM enable such an investigation forthe first time in a distant elliptical, which can offer valuableinsight into late-time feedback in massive elliptical galaxies.Specifically, now that we have robustly detected highlyionized gas in the lens galaxy and constrained its properties,we can calculate the mass budget in the warm ( T ∼ K)gas phase and compare it to the previously estimated massbudget in the cool ISM (Z17). For the cool phase, Z17 esti-mated a total Fe mass of M Fe ∼ (5 − × ( f c , cool ) M (cid:12) at d < kpc ( ≈ r e , matching the region probed by the doublylensed QSO), where f c , cool is the cool gas covering fraction.The corresponding total mass in the cool phase is M cool ∼ (4 − × (cid:18) f c , cool . (cid:19)(cid:18) Z cool Z (cid:12) (cid:19) − M (cid:12) , (1)where Z cool is the cool gas metallicity. For f c , cool ≈ and asolar metallicity gas, which Z17 inferred for the cool phase,the inferred mass in the cool ( T ∼ K) ISM is M cool ∼ (4 − × M (cid:12) .Assuming that the observed N (O VI ) along sightline A isrepresentative at d < kpc, the estimated mass in the O VI -bearing phase of the ISM is M warm ∼ × (cid:18) f c , warm . (cid:19)(cid:18) Z warm Z (cid:12) (cid:19) − (cid:18) f O . (cid:19) − M (cid:12) , (2)where f c , warm is the covering fraction of the warm phase, Z warm is its metallicity, and f O is the ionization fraction of O ions. If we further assume a unity covering fraction andsolar gas metallicity for the warm ( T ∼ K) ISM phase,and adopt a reasonable f O ≈ . − . which is predicted for a wide range of physical conditions at T ∼ K (e.g.,Oppenheimer & Schaye 2013), we find a total gas mass of M warm ∼ (1 . − × M (cid:12) in the warm ISM phase thatis likely traced by O VI absorption. Despite the uncertain-ties inherent in our simple calculations, the estimated massbudgets in the cool and warm ISM phases are comparable towithin a factor of a few if the two phases have similar gascovering fractions and metallicities.In the physical picture where the observed O VI absorbertraces transitional temperature ( T ∼ K) gas that is ra-diatively cooling from a virialized hot phase ( T ∼ K), M warm is proportional to the mass flow rate into the coolISM following ˙ M cool = M warm /t cool , where t cool is thecooling timescale of the O VI -bearing gas. The coolingtimescale depends on the gas temperature, metallicity, anddensity. For T ≈ . K and a solar-metallicity gas witha density of n H = 10 − cm − , typical of the hot halo ofmassive ellipticals at d ≈ kpc (e.g., Singh et al. 2018),the expected cooling time is t cool ≈ − Myr (Gnat &Sternberg 2007; Oppenheimer & Schaye 2013) with a totalcooling rate of ∼ (1 − × erg yr − . Thus, in a ra-diative cooling scenario the estimated M warm translates toa mass flow rate of ˙ M cool ∼ . − . (cid:12) yr − at d < kpc. If the bulk of this flow cooled to T (cid:46) K and re-mained in this phase, we should expect M cool (cid:29) M warm over a timescale of ∼ Myr in the absence of significantstar-formation activity. Considering Z16 found a minimumstellar population age of > Gyr and no detectable star for-mation (
SFR < . (cid:12) yr − ) in the lens galaxy, this calcu-lation suggests that most of the cooling gas is reheated to thecoronal phase. To heat the gas back to virial temperature,the required heating rate is ˙ E heat ∼ (1 − × erg yr − ,assuming T vir ≈ × K given the estimated mass of thedark-matter host halo of the lens (Z16).Observations of nearby massive ellipticals show that me-chanical feedback (often dubbed “radio-mode feedback”)from active galactic nuclei (AGNs) can output as much poweras ˙ E AGN ∼ − erg yr − (e.g., Werner et al. 2019).If the lens galaxy of HE 0047 − VI -traced cooling gas. There are twocaveats to this statement, however. Because the estimatedcooling time of a T ∼ K gas is short ( ∼ yr) owingto the expected high gas densities at d < kpc, the actualamount of available heating depends sensitively on the radio-mode duty cycle (i.e., the fraction of time that an AGN isin radio mode). The radio-mode duty cycle in ellipticals hasbeen estimated to be no more than ≈ outside of richcluster environments (O’Sullivan et al. 2017). Furthermore,even if an AGN is currently on, its energy output is likelyto be distributed over a large volume in the gaseous halo.Indeed, observations of large X-ray cavities/bubbles and ex- In principle, the cool gas could also be depleted primarily by furthercooling into the cold ( T (cid:46) K) phase probed by H molecules. However,we consider this scenario unlikely given the inferred low mass fraction ofmolecular gas in the lens ISM, which is also consistent with observations ofnearby ellipticals (§4.2). ULTIPHASE GAS IN A MASSIVE ELLIPTICAL GALAXY AT z = 0 . ∼ kpc or larger in the hot halo (e.g., McNamara & Nulsen 2007;Fabian 2012). In conclusion, whether AGNs are viable as acontinuous heating source requires not only a high duty cyclebut also that its mechanical energy can be effectively coupledwith the ISM on ∼ kpc scales in the galaxy.Alternatively, we consider heating sources associated withthe old stellar populations themselves. Previous analytic andsimulation studies suggest that feedback from SNe Ia andstellar winds from asymptotic giant branch (AGB) stars mayoffset radiative cooling from diffuse gas in massive ellipticalgalaxies with M star ≈ M (cid:12) (e.g., Conroy et al. 2015; Liet al. 2018). Empirically, the observed high [Fe / Mg] abun-dance ratios at d (cid:46) kpc from quiescent galaxies (Z16;Zahedy et al. 2017a) also supports the idea that their ISM hasbeen subjected to significant influence from recent SNe Ia.Using the mean SN Ia rate in nearby ellipticals (e.g., Man-nucci et al. 2005), Z17 estimated an integrated SN Ia rate of ∼ . per century within d < kpc from the massive ellip-tical lens galaxy. Multiplying this rate by a mean energy of erg per SN Ia, we estimate that the heating rate availablefrom SNe Ia is ˙ E Ia ∼ × erg yr − , which is comparableto the required heating.In addition to SNe Ia heating, Conroy et al. (2015) alsoconsidered how materials ejected from AGB stars can in-teract with and heat the ambient ISM in elliptical galaxies.Using their analytic formula for AGB heating rate, we es-timate a heating rate of ˙ E AGB ∼ × erg yr − in thelens galaxy of HE 0047 − VI absorption andprevent a large accumulation of cold gas in the ISM, even inthe absence of strong feedback from an active nucleus. CONCLUSIONSOur analysis of the medium-resolution FUV spectrum oflensed QSO sightline HE 0047 − A has revealed a com-plex, multiphase gas at d = 4 . kpc from the lens andyielded the first constraints on multiphase ISM propertiesin a massive quiescent galaxy ( M star ≈ M (cid:12) ) beyondthe local Universe. H gas is detected with column density log N (H ) / cm − = 17 . +0 . − . and a molecular gas fractionof f H = 2 − . Furthermore, the ISM exhibits O VI and N V absorptions with a distinct kinematic structure fromthat of the low ions (e.g. Mg II ; Z16), indicating that thesehigh ions arise in a different gas phase. The highly ionizedphase has a total log N (O VI ) / cm − = 15 . ± . and log N (N V ) / cm − = 14 . ± . , among the strongest associ-ated with z < galaxies. The low- and high-ionization gasphases are spatially extended on ∼ kpc scale, which is incontrast to the patchier H spatial distribution on this scale.We have investigated how the ISM is partitioned by massinto its different phases and examined its implications onlate-time feedback in the galaxy. Specifically, the mass in thehighly ionized ISM phase is M warm ∼ (1 . − × M (cid:12) at d < kpc, comparable to the estimated mass in the cool( T (cid:46) K) ISM. Assuming the high-ionization gas orig-inates in a transient warm ( T ∼ K) phase undergoingradiative cooling from a hot halo surrounding the galaxy, theinferred mass accretion rate is ∼ . − . (cid:12) yr − . Thelack of star-formation activity ( SFR < . (cid:12) yr − ) in thegalaxy suggests that most of this flow is reheated to the hotphase, at a rate of ˙ E heat ∼ (1 − × erg yr − . Con-tinuous heating from evolved stellar populations (primarilySNe Ia but also AGB winds) in the massive elliptical galaxymay suffice to prevent a large accumulation of cold gas in theISM, even in the absence of strong AGN feedback. Whilethis conclusion is based on a single galaxy, our study under-scores the important role that evolved stellar populations canplay in maintaining the low star-formation rate in massivequiescent galaxies over cosmic time.The authors thank the anonymous referee for thoughtfulcomments that helped improve the presentation of this pa-per. We thank Patrick Petitjean for providing his H linelist, and Sean Johnson and Ben Rosenwasser for insightfuldiscussions. FSZ acknowledges support of a Carnegie Fel-lowship from the Observatories of the Carnegie Institutionfor Science. FSZ and HWC acknowledge partial supportfrom HST-GO-15250.004A. FSZ, HWC, and EB acknowl-edge partial support from HST-GO-15163.001A and NSFAST-1715692 grants. This work is based on data gatheredwith the NASA/ESA Hubble Space Telescope operated bythe Space Telescope Science Institute and the Association ofUniversities for Research in Astronomy, Inc., under NASAcontract NAS 5-26555. Additional data shown here weregathered with the 6.5 m Magellan Telescopes located at LasCampanas Observatory in Chile.REFERENCES
Albornoz V´asquez, D., Rahmani, H., Noterdaeme, P., et al. 2014, A&A,562, A88Balashev, S. A., Klimenko, V. V., Noterdaeme, P., et al. 2019, MNRAS,490, 2668Black, J. H. 1987, Interstellar Processes, 731Boettcher, E., Chen, H.-W., Zahedy, F. S., et al. 2020, arXiv:2010.11958Bowen, D. V., & Chelouche, D. 2011, ApJ, 727, 47Chen, H.-W., Helsby, J. E., Gauthier, J.-R., et al. 2010, ApJ, 714, 1521Chen, H.-W. 2017a, Gas Accretion onto Galaxies, 167Chen, H.-W. 2017b, Outskirts of Galaxies, 291Chen, H.-W., Zahedy, F. S., Johnson, S. D., et al. 2018, MNRAS, 479, 2547 Conroy, C., van Dokkum, P. G., & Kravtsov, A. 2015, ApJ, 803, 77Crighton, N. H. M., Bechtold, J., Carswell, R. F., et al. 2013, MNRAS, 433,178Davis, T. A., Alatalo, K., Bureau, M., et al. 2013, MNRAS, 429, 534.doi:10.1093/mnras/sts353Fabian, A. C. 2012, ARA&A, 50, 455Ferland, G. J., Porter, R. L., van Hoof, P. A. M., et al. 2013, Rev. MexicanaAstron. Astrofis., 49, 137Foreman-Mackey, D., Hogg, D. W., Lang, D., et al. 2013, PASP, 125, 306Gauthier, J.-R., Chen, H.-W., & Tinker, J. L. 2009, ApJ, 702, 50Gauthier, J.-R., Chen, H.-W., & Tinker, J. L. 2010, ApJ, 716, 1263
AHEDY ET AL ..